I really, really like starlink. I think it is the most usefull of Musk ideas, but my opinion is tainted because i love sailing, and i have a transatlantic in a year and a half. I only skimmed this, but anybody else thought the papers were weird? lot of photograph, not a lot of details on the plan.
I have two big red flags personally.
- The first one can easily be understood by reading my previous comments on OpenAI. AI addressable market is 30 trillions? what? did they compare chineese free/cheap models to grok? or even to Anthropic/OpenAI frontier models? This is delusional. Maybe some devs doing heavy math will have a benefit using frontier models right now, but at this very moment i don't see why i should pay for AI for my daily usage (and i was an early subscriber of chatGPT) since free models are so good (i'm not talking about code here). Honestly i'm ready to pay 20€/month at most atm, and the number is declining fast as the chineese models got better.
- Point-to-point travel. No, it won't work. Even without taking into account the locations (which will have to be very remote) and the danger, you don't go in a 0g plane without a thorough medical exam, so i'm pretty sure it will be worse to get in a rocket. Also, fuel efficiency on the new Safran/GE RISE aim for 20% efficiency gain for the new airplane generation, which will make very hard to be competitive with air travel for any rocket-powered system. I want a _very_ detailed plan if you want to make me believe you have thought about all the issues, and this just isn't this.
I find the TAM numbers inspiring. I think we need a new name for it, like speculative TAM, or "STAM" as in "once we capture 5% of STAM for orbital data centers, our EBITDA will be $10 quadrillion."
I really, really like starlink. I think it is the most usefull of Musk ideas, but my opinion is tainted because i love sailing, and i have a transatlantic in a year and a half. I only skimmed this, but anybody else thought the papers were weird? lot of photograph, not a lot of details on the plan.
I have two big red flags personally.
- The first one can easily be understood by reading my previous comments on OpenAI. AI addressable market is 30 trillions? what? did they compare chineese free/cheap models to grok? or even to Anthropic/OpenAI frontier models? This is delusional. Maybe some devs doing heavy math will have a benefit using frontier models right now, but at this very moment i don't see why i should pay for AI for my daily usage (and i was an early subscriber of chatGPT) since free models are so good (i'm not talking about code here). Honestly i'm ready to pay 20€/month at most atm, and the number is declining fast as the chineese models got better.
- Point-to-point travel. No, it won't work. Even without taking into account the locations (which will have to be very remote) and the danger, you don't go in a 0g plane without a thorough medical exam, so i'm pretty sure it will be worse to get in a rocket. Also, fuel efficiency on the new Safran/GE RISE aim for 20% efficiency gain for the new airplane generation, which will make very hard to be competitive with air travel for any rocket-powered system. I want a _very_ detailed plan if you want to make me believe you have thought about all the issues, and this just isn't this.
I find the TAM numbers inspiring. I think we need a new name for it, like speculative TAM, or "STAM" as in "once we capture 5% of STAM for orbital data centers, our EBITDA will be $10 quadrillion."